Trailed record inverts case class field order

[wzorgdrager]

After a case class has been transformed to a (Trailed?)Record the fields (from the case class) are inverted.

Example:
case class Test(field1 : String, field2: String, field3: String)
Currently, if this case class is transformed to a TrailedRecord I need to call record.getField(0) to get field3, record.getField(1) to get field2 and record.getField(2) to get field3.

I guess it would be preferable to have it in the same ordering as the case class.

[nvankaam]

Strange, I hope then that classtags in scala provides any guarantee on the order of fields.
In what case however do you directly have to call getField with an index? All fields should be named, so you should always be able to get your value using a column name.

[wzorgdrager]

I'm using the KafkaRowSource so I can't get fields by column names. I probably even want to convert the Kafka topic outputs into their original case classes again, but I'm not sure if that is possible (I can also re-parse it into a case class)?

[nvankaam]

It is possible to just get the case classes from Kafka again. If you use a "TrailedRecordSource"you can call the "SubjectFactoryController" to get an "untransformer" which transforms a trailed record back into a generic type.

I did not add a KafkaSource that directly exposes the case classes because using them instead of the trailed record would drop context information from the record's "trail". This trail was mainly important for my idea about how to implement the query language. Since we no longer want our own query language (but use flink), i think we no longer need that trail and could expose the case classes from the kafka sources directly.

[wzorgdrager]

Aha, I didn't know that. Thanks for the heads up!